Mustafa Suleyman: MAI Transcribe 2 Is 10x Faster Than GPT Transcribe
Try MAI-Transcribe-2 the fastest, cheapest and most accurate transcription model in the world! 10x faster than GPT‑Transcribe 5x faster than Gemini 3.5 $0.10 per hour of audio
Context
A cheap, fast transcription model would matter for anyone who needs to turn large amounts of audio into text, but this post is a promo of a model Microsoft already launched.
Mustafa Suleyman, the CEO of Microsoft AI, posted that people should try MAI Transcribe 2, which he called the fastest, cheapest, and most accurate transcription model in the world. He said it is 10x faster than GPT Transcribe, 5x faster than Gemini 3.5, and costs $0.10 per hour of audio. Microsoft AI first published those claims in a September 3, 2026 article introducing the model.
Deep dive
Mustafa Suleyman posted today asking people to try MAI Transcribe 2, which he called the fastest, cheapest, and most accurate transcription model in the world. He wrote that it is 10x faster than GPT Transcribe, 5x faster than Gemini 3.5, and priced at $0.10 per hour of audio.
Those claims are not a new launch today. Microsoft AI published an article on September 3, 2026, introducing MAI Transcribe 2 with the same speed and price figures. That article said that, based on evals run by Artificial Analysis, the model is 10x faster than OpenAI's GPT Transcribe, 7x faster than ElevenLabs' Scribe v2, and 5x faster than Gemini 3.5 Transcribe, while delivering higher accuracy. It also said the model ranks first on the FLEURS benchmark across 60 languages with an average word error rate of 5.2%, and second on the Artificial Analysis word error rate leaderboard.
The September article said the $0.10 per hour price was a limited time offer at launch until the end of the year, and that the model could be demoed through Microsoft Foundry, MAI Playground, and Open Router. Suleyman's post today does not say whether that price is still an introductory offer or whether anything about the model changed on Nov 10. The speed, accuracy, and price figures in the post should be read as Microsoft's own claims, supported in the earlier article by Artificial Analysis evals that were not independently rechecked in this scan.
What to expect next
The useful check is whether the $0.10 introductory price is still in effect through the end of 2026, and whether later Microsoft posts add a new version or only repeat the September claims. Independent benches after today will show whether the speed and accuracy lead still holds against current GPT Transcribe and Gemini models. Until Microsoft posts a dated change, this should be treated as a promo of a model that has been public since September 3.